https://orcid.org/0000-0003-4251-2090
ABSTRACT
Water rights trading policy is an important economic tool in China's water resource management. Accurately estimating the level of water resource utilization efficiency in experimental areas and impartially assessing the impact of the water rights trading market's operation on water resource utilization efficiency have policy implications for improving water resource governance tools and capabilities. Based on the panel data of 19 provinces (regions) in China from 2010 to 2019, this paper describes the development of a three-stage data envelopment analysis model to measure water use efficiency and the application of the difference-in-differences method based on propensity score matching (PSM-DID) to examine the effect of water rights trading policies on the model. The results indicate that (1) after the introduction of the water rights trading policy, the water use efficiency in the experimental areas improved substantially; (2) the PSM-DID method also confirms this finding, and water rights trading has a positive influence on water use efficiency. The research outcomes of this study can be used to objectively appraise the policy performance of the water rights trading policy and provide a theoretical foundation for its nationwide implementation.
HIGHLIGHTS
The three-stage data envelopment analysis (DEA) model is used to calculate the water resources utilization efficiency (WRUE), which overcomes the shortcomings of the traditional DEA model ignoring environmental factors and uncontrollable factors and improves the scientificity and accuracy of the calculation.
The difference-in-differences method based on propensity score matching is used to compare the changes in WRUE in the pilot provinces and non-pilot provinces before and after implementation of the policy.
INTRODUCTION
China is rich in water resources, but the distribution of water resources is quite different, showing a trend of ‘more in the south and less in the north’, and the per capita water consumption is relatively low, which hinders China's green and sustainable development (Dong et al. 2019). To this end, China has established a water resources planning system, a water intake permit system, a planned water use system, and so on, to improve the efficiency of water resources utilization through institutionalized means (Shen & Speed 2009). However, the model of restricting water resources management only through administrative means is no longer suitable for the needs of Chinese societal development, and problems such as high water resources management costs, low management efficiency, and fixed water resources allocation have become increasingly prominent (Tian et al. 2020). Therefore, in 2014, the Chinese government implemented the water rights (WRs) trading policy and carried out water rights pilot projects in seven provinces and regions, namely, Henan, Ningxia, Jiangxi, Hubei, Inner Mongolia, Gansu, and Guangdong, trying to optimize the management mode and supply and demand structure of water resources utilization by market-oriented means to maximize the efficiency of water resources utilization (Tian et al. 2022). This raises the following questions: Can the WRs trading policy improve the efficiency of water resources utilization? Can the WRs trading policy be promoted and popularized across the country?
The map of the study area.
China's water resources management policy has undergone a process from government intervention to market regulation (Dou et al. 2014). This can be roughly divided into stages such as public welfare free water use, policy-based low-price water supply, water supply cost accounting, and market-oriented water supply management (Li et al. 2020).
(1) In the 1950s, the government led the development of water resources, and China was basically at the stage of free water use.
(2) The 1965 ‘Trial Measures for the Collection, Use and Management of Water Fees for Reservoir Projects’ marked the end of the stage of free water use, and China tried to implement the water fee system. However, due to the influence of traditional concepts, this was difficult to implement (Wang et al. 2019a). In 1985, the ‘Water Fee Measures’ were promulgated to end the stage of low-price water use, and the charges were approved separately according to the differences in water resources. Through the promulgation of the ‘Water Law’ in 1988, water resources fees began to be fully collected (Chou et al. 2021), and the commodity nature of water resources gradually emerged.
(3) In 2014, the Ministry of Water Resources issued the ‘Guiding Opinions on Deepening Water Conservancy Reform’, which established the key position of the market mechanism in the utilization of water resources (Sun et al. 2023), and implemented measures to allocate water resources based on market specifications, market prices, and market competition. Meanwhile, the Ministry of Water Resources, taking into account the sustainable use of water resources, has set regional water consumption quotas, which has also created conditions for the formation of WRs trading markets. In 2016, the China Water Exchange was officially established in Beijing to guide WRs trading and provide consultation, evaluation, and other services to facilitate water trading practices. At the same time, the form, price, and transaction information of WRs transactions in various regions are disclosed, to supervise WRs transactions in China. In 2022, the ‘Guiding Opinions on Promoting the Reform of the Right to Use Water’ was promulgated, reaffirming support for advancing WRs reform. It underscores the necessity of accelerating the initial allocation of WRs, expediting market-based WRs trading, refining and optimizing WRs trading platforms, and intensifying oversight over water resources transactions.
The ultimate goal of WRs trading is to save water by improving the efficiency of water resources utilization. However, the current WRs trading practice lacks consideration of water quality, which may cause contractual disputes between buyers and sellers (Shen et al. 2020). Therefore, on the one hand, it is imperative to clarify the property rights of water resources, curtail excessive water demand through restricted and indexed water use standards, and enhance the efficiency of water use via a strong binding system (Du et al. 2022). On the other hand, it is necessary to further improve the price mechanism of WRs trading and fully consider factors such as water quality and quantity based on the opportunity cost and the actual cost of water use. Thus, the water value concept and action logic of ‘acquiring according to demand and requesting according to price’ are gradually formed to improve the efficiency of water resources utilization (Zhu & Fang 2022).
However, some scholars believe that the WRs trading policy negatively affects the improvement of water resources utilization efficiency. The main reasons include the following:
The government is the regulator of the WRs trading policy, but the phenomenon of government intervention in pricing occurs in the process of policy practice (Su & Zhang 2022).
The rationality of the transaction price of WRs is doubtful, there are ‘policy-based low prices’, and there is an unreasonable phenomenon of ‘the more water is used, the more cost-effective it is’ (Chou et al. 2021).
The ownership of regional WRs is vague, leading to the ‘tragedy of public water’ that cannot be investigated for excessive water use (Gu et al. 2022).
As one of the basic elements of life and production, water resources also affect the effective allocation of supply chains. The WRs trading policy can further realize the model optimization of the water resources supply chain and the product supply chain formed by relying on water resources. On the one hand, improvement of the WRs trading policy can ensure normal circulation in the water resources supply chain (Wang 2022). It can also optimize the response allocation between the government and the supply and demand of WRs and strengthen the response ability to deal with the gaps in water resources, as well as reducing supply chain operating costs (Li et al. 2021), promoting a double cycle of water resources and capital circulation and improving water resources utilization efficiency (Qin 2017). On the other hand, responding to emergencies and optimizing the supply chain configuration of water-dependent products will further promote the improvement of water resources technical efficiency.
Disruptions to economic trading activities such as the outbreak of COVID-19 in 2019, blocked China's maritime supply chain (March et al. 2021) and increased freight costs (Michail & Melas 2020). On the one hand, the blocked supply chain leads to unsalable products, which reduces the efficiency of regional water resources utilization. On the other hand, the obstruction of product imports leads to the consumption of more water resources for product manufacturing. In the face of such problems, we can only seek to improve the efficiency of water resources utilization by optimizing the local water resources supply chain and improving the technical efficiency of water resources.
A review of previous studies found that, theoretically speaking, WRs trading is an important economic means which can promote the conservation, protection, optimal allocation, and efficient use of water resources. However, water resources have more complex natural and social attributes than other natural resources. Whether the WRs trading policy can genuinely exert a role in enhancing the efficiency of regional water resources utilization requires further elaboration.
RESEARCH METHODS AND DATA SOURCES
Research methods
Three-stage DEA model
Water resources utilization efficiency refers to the ratio of the economic and environmental output obtained using water resources to calculate the inputs of labor, capital, water resources, and other production factors (Zhao et al. 2022). DEA is a nonparametric method for measuring the operational efficiency of multi-input and multi-output economic systems that has been widely used in efficiency measurement research (Gautam et al. 2020). The classic DEA method does not accurately reflect the real efficiency level of decision-making units because it disregards the influence of external environmental factors and stochastic disturbances on its appraisal of efficiency. To assure the correctness of the results, in this research, the three-stage DEA model was adopted to evaluate each province's water resources utilization efficiency over time. The particular procedures are as follows.
Stage 1: the traditional DEA model
, there is room for improvement in production. When the efficiency value is 1, the production is efficient. This study investigated how to reduce water resources consumption while maintaining output in the context of increasing water scarcity. As a result, the input-oriented DEA model, namely, the BCC model, is used for analysis in this article. This boils down to a linear programming issue, which may be written as
and 
denote, respectively, the input and output vectors of the j decision-making unit. 
represents the weighting coefficient of the i input and the r output. 
represents the efficiency value of the j decision-making unit, and its range is between 0 and 1; the greater the value, the more efficient the operation. The efficiency value computed by the BCC model represents the comprehensive technical efficiency, which is the water resources utilization efficiency in this study. This can be further disaggregated into pure technical efficiency (PTE) and scale efficiency (SE).Stage 2: stochastic Frontier approach
is the slack variable associated with the i input variable of the j decision-making unit. 
is the environment variable, and 
represents the impact of the environment variable on the slack variable 
. 
is the parameter vector to be estimated. 
represents the mixed error term, where 
represents the random error term, that is, the interference of external random factors on the input slack variable. It is generally assumed that 
is the managerial inefficiency term, that is, the effect of managerial factors on input slack variables. To eliminate the impact of external environmental factors and stochastic noise, the inputs of relatively inefficient decision-making units are adjusted one by one according to the following formula:
In the equation, 
and 
represent the input value of each decision-making unit before adjustment and after adjustment, respectively. 
denotes adjusting with the external environment to eliminate the influence. 
denotes eliminating the influence of noise.
Stage 3: water use efficiency is recalculated using new input variables
We replace the original input variable value 
with the adjusted input variable value 
, keep the output variable constant, and run according to the first-stage model again to obtain a new estimated value of water resources utilization efficiency. The new efficiency estimates, which are obtained by eliminating the effects of external environmental factors and random noise, accurately represent the water resources utilization efficiency of each region.
When evaluating water resources utilization efficiency, it is first necessary to construct a scientific and reasonable evaluation index system. In this study, based on the existing literature and the correlation analysis results of various indicators, the indicators with a high degree of linear correlation are eliminated, while considering the overall relevance of the evaluation indicators and data availability, to build a systematic and comprehensive evaluation index system. The specific meaning of each evaluation index is as follows:
(1) Total investment in fixed assets, which is expressed in units of 100 million yuan. Although WRs are material resources that human beings rely on for survival and production, economic benefits can only be generated when they are combined with other means of production. As an important measure of the total investment in societal resources, the investment in fixed assets by the entire society can reflect the country's efforts to promote economic development and the construction of water resources governance. Therefore, the total investment in fixed assets is used as the input index for the evaluation of water resources utilization efficiency.
(2) Permanent population, which is expressed in units of 10,000 people. Due to the continuous flow of population between regions, the resident population is the main body of water consumption in a certain region. In addition, the resident population is an important basis for evaluating the production capacity of the national economy. Therefore, the regional resident population is selected as the input index for the evaluation of water resources utilization efficiency. The resident population of each region here is measured by the statistical population of each region.
(3) Total water consumption, which is expressed in units of 100 million m3. Only when water resources consumption occurs can there be calculation of water resources utilization efficiency. Therefore, the total water consumption is used as the input index for the evaluation of water resources utilization efficiency.
(4) Regional gross product, which is expressed in units of 100 million yuan. The consumption of WRs needs to meet the requirements of human survival and production and create social and economic value through the use of WRs. Therefore, the GDP of each region is used as the output index for the evaluation of water resources utilization efficiency.
Environmental variables mainly refer to the factors that influence water resources utilization efficiency and cannot be subjectively resolved in a short time span. This paper mainly describes the following three types of environmental variables and their impacts on water resources utilization efficiency:
(1) Industrial structure. There is a close relationship between industrial structure and water resources utilization, among which agricultural water consumption accounts for a large proportion, exceeding 60% in most provinces, followed by industrial water, accounting for approximately 24% of the total water usage. However, due to the extensive use of agricultural irrigation water, its proportion in the entire industrial system has a greater impact on water resources utilization efficiency. As a result, the proportion of the main industry's added value in GDP is used in this article to quantify the industrial structure and consider its impact on water resources utilization efficiency.
(2) Water resources endowment. Compared with resource-poor areas, it is easier to obtain resources in resource-rich areas, thus ignoring the sustainable use of resources, resulting in low resource utilization, and then creating the ‘resource curse’ problem. Following most of the literature and industry habits, per capita water resources is used to measure the water resources endowment conditions in various regions.
(3) Level of economic development. Compared with economically undeveloped areas, economically developed areas will further optimize the allocation of water resources, thereby promoting the improvement of water resources utilization efficiency. As a result, per capita GDP is used to assess each region's economic development level.
PSM-DID method
DID is a commonly used econometric method in policy analysis and evaluation. By dividing the experimental samples into a ‘treatment group’ and ‘control group’ according to whether they are affected by the policy, and comparing and analyzing the changes in the dependent variables of the two groups of samples before and after the implementation of the policy, the ‘net impact’ of the policy on the dependent variable can be obtained (Yang et al. 2021; Zhi et al. 2022).
The subscripts i and and t represent the province and time, respectively. 
is the explained variable of province i in year t, and this paper refers to the water resources utilization efficiency of province i in year t. 
and 
are time and region dummy variables set. 
is a constant term. 
is the coefficient of the time dummy variable 
. 
is the coefficient of the regional dummy variable 
. 
is the coefficient of the interaction term 
of the time dummy variable and the region dummy variable. 
is important in the whole model as it represents the implementation effect of the policy; if its value is significantly positive, this indicates that WRs trading can promote the improvement of water resources utilization efficiency.
For the policy pilot time dummy variable (
), before the implementation of the WRs trading policy (2010–2013), 
. After the implementation of the WRs trading policy (2014–2019), 
. For the dummy variable (
) of policy pilot areas, 7 provinces (regions) from the national pilot program are used for analysis, as well as 12 other provinces (regions) that did not carry out the WRs pilot program. 
is the interaction term of the dummy variables 
and 
, indicating the net impact before and after the implementation of the policy. 
is a series of control variables that may affect the efficiency of water resources utilization. Based on previous studies, this paper selects eight covariates, namely the proportion of the secondary industry, proportion of the tertiary industry, urbanization rate, population density, population quality, investment in fixed assets for sewage treatment, area of water-saving irrigation, and investment in environmental pollution control, to test the matching effect of the samples. The definitions of each variable are shown in Table 1.
Variable definition table
| Variable name | Variable code | Variable definitions |
|---|---|---|
| Dependent variable | ||
| water resources utilization efficiency | y | Based on three-stage DEA model calculation |
| Core explanatory variable | ||
| Policy pilot areas |  |  pilot provinces that have not carried out WRs trading  pilot provinces for WRs trading |
| Policy pilot time |  |  before the implementation of the WRs trading policy  after the implementation of the WRs trading policy |
| Double-differenced term | DID |  |
| Control variable | ||
| Proportion of the secondary industry | x1 | Output value of the secondary industry (GDP) |
| Proportion of the tertiary industry | x2 | Output value of the tertiary industry (GDP) |
| Urbanization rate | x3 | Year-end urban population/total population of the region |
| Population density | x4 | Total population/urban land area |
| Population quality | x5 | Average years of education |
| Investment in fixed assets for sewage treatment | x6 | 10,000 yuan |
| Area of water-saving irrigation | x7 | 1,000 ha |
| Investment in environmental pollution control | x8 | Environmental pollution control investment (GDP) |
| Variable name | Variable code | Variable definitions |
|---|---|---|
| Dependent variable | ||
| water resources utilization efficiency | y | Based on three-stage DEA model calculation |
| Core explanatory variable | ||
| Policy pilot areas | | pilot provinces that have not carried out WRs trading pilot provinces for WRs trading |
| Policy pilot time | | before the implementation of the WRs trading policy after the implementation of the WRs trading policy |
| Double-differenced term | DID | |
| Control variable | ||
| Proportion of the secondary industry | x1 | Output value of the secondary industry (GDP) |
| Proportion of the tertiary industry | x2 | Output value of the tertiary industry (GDP) |
| Urbanization rate | x3 | Year-end urban population/total population of the region |
| Population density | x4 | Total population/urban land area |
| Population quality | x5 | Average years of education |
| Investment in fixed assets for sewage treatment | x6 | 10,000 yuan |
| Area of water-saving irrigation | x7 | 1,000 ha |
| Investment in environmental pollution control | x8 | Environmental pollution control investment (GDP) |
Then, kernel matching was employed for balance testing, and the results are presented in Table 8. It can be seen from Table 2 that the absolute value of the deviation of each variable is significantly less than 20% after matching, indicating that the matching process was effective (Rosenbaum & Rubin 1983). At the same time, there was no significant difference in T-test statistics, suggesting that there was no systematic difference between the treatment and control groups, and there was no bias in the selection of samples, so the application of DID based on PSM was supported.
Balance test of PSM
| Variable | Not matched (U) | Average | Deviation % | Reduction rate of deviation before and after matching % | t-test | ||
|---|---|---|---|---|---|---|---|
| After matching (M) | Treatment group | Control group | t | p > |t| | |||
| x1 | U | 3.878 | 3.750 | 74.600 | 92.500 | 4.000 | 0.000 |
| M | 3.876 | 3.867 | 5.600 | 0.330 | 0.742 | ||
| x2 | U | 3.706 | 3.790 | −51.900 | 71.400 | −2.920 | 0.004 |
| M | 3.719 | 3.743 | −14.900 | −0.600 | 0.550 | ||
| x3 | U | 3.948 | 3.985 | −16.900 | 39.400 | −0.930 | 0.355 |
| M | 3.983 | 4.005 | −10.200 | −0.390 | 0.695 | ||
| x4 | U | 7.932 | 7.885 | 10.300 | 72.300 | 0.620 | 0.534 |
| M | 7.877 | 7.890 | −2.800 | −0.120 | 0.905 | ||
| x5 | U | 2.248 | 2.247 | 2.200 | − 473.400 | 0.120 | 0.906 |
| M | 2.250 | 2.257 | −12.800 | −0.530 | 0.597 | ||
| x6 | U | 11.245 | 10.873 | 29.000 | 88.200 | 1.660 | 0.099 |
| M | 11.189 | 11.145 | 3.400 | 0.150 | 0.882 | ||
| x7 | U | 6.517 | 6.277 | 25.300 | 24.800 | 1.410 | 0.160 |
| M | 6.500 | 6.320 | 19.100 | 0.850 | 0.397 | ||
| x8 | U | 1.618 | 1.224 | 56.600 | 76.500 | 3.470 | 0.001 |
| M | 1.482 | 1.390 | 13.300 | 0.550 | 0.582 | ||
| Variable | Not matched (U) | Average | Deviation % | Reduction rate of deviation before and after matching % | t-test | ||
|---|---|---|---|---|---|---|---|
| After matching (M) | Treatment group | Control group | t | p > |t| | |||
| x1 | U | 3.878 | 3.750 | 74.600 | 92.500 | 4.000 | 0.000 |
| M | 3.876 | 3.867 | 5.600 | 0.330 | 0.742 | ||
| x2 | U | 3.706 | 3.790 | −51.900 | 71.400 | −2.920 | 0.004 |
| M | 3.719 | 3.743 | −14.900 | −0.600 | 0.550 | ||
| x3 | U | 3.948 | 3.985 | −16.900 | 39.400 | −0.930 | 0.355 |
| M | 3.983 | 4.005 | −10.200 | −0.390 | 0.695 | ||
| x4 | U | 7.932 | 7.885 | 10.300 | 72.300 | 0.620 | 0.534 |
| M | 7.877 | 7.890 | −2.800 | −0.120 | 0.905 | ||
| x5 | U | 2.248 | 2.247 | 2.200 | − 473.400 | 0.120 | 0.906 |
| M | 2.250 | 2.257 | −12.800 | −0.530 | 0.597 | ||
| x6 | U | 11.245 | 10.873 | 29.000 | 88.200 | 1.660 | 0.099 |
| M | 11.189 | 11.145 | 3.400 | 0.150 | 0.882 | ||
| x7 | U | 6.517 | 6.277 | 25.300 | 24.800 | 1.410 | 0.160 |
| M | 6.500 | 6.320 | 19.100 | 0.850 | 0.397 | ||
| x8 | U | 1.618 | 1.224 | 56.600 | 76.500 | 3.470 | 0.001 |
| M | 1.482 | 1.390 | 13.300 | 0.550 | 0.582 | ||
Data source and processing
The above research data were obtained from the ‘China Statistical Yearbook’ from 2011 to 2020 and the ‘water resources Bulletin’ of various regions over the years. Before the calculation of water resources utilization efficiency, SPSS Statistics was used to perform descriptive statistical analyses on the collected data, and Pearson's correlation test and a multicollinearity test were performed. The running results are displayed in Tables 3–5.
Descriptive statistics of input–output indicators
| Observation quantity | Minimum | Maximum | Average | Standard deviation | |
|---|---|---|---|---|---|
| Regional gross product | 152 | 1,350.430 | 89,705.230 | 20,082.220 | 18,278.530 |
| Total investment in fixed asset | 152 | 1,016.900 | 53,277.000 | 14,103.260 | 10,808.930 |
| Permanent population | 152 | 563.000 | 11,169.000 | 4,397.860 | 2,921.130 |
| Total water consumption | 152 | 22.490 | 591.300 | 205.190 | 145.180 |
| Effective quantity | 152 |
| Observation quantity | Minimum | Maximum | Average | Standard deviation | |
|---|---|---|---|---|---|
| Regional gross product | 152 | 1,350.430 | 89,705.230 | 20,082.220 | 18,278.530 |
| Total investment in fixed asset | 152 | 1,016.900 | 53,277.000 | 14,103.260 | 10,808.930 |
| Permanent population | 152 | 563.000 | 11,169.000 | 4,397.860 | 2,921.130 |
| Total water consumption | 152 | 22.490 | 591.300 | 205.190 | 145.180 |
| Effective quantity | 152 |
Pearson correlation test result table
| Regional gross product | |
|---|---|
| Total investment in fixed asset | 0.857a |
| Permanent population | 0.800a |
| Total water consumption | 0.779a |
| Regional gross product | |
|---|---|
| Total investment in fixed asset | 0.857a |
| Permanent population | 0.800a |
| Total water consumption | 0.779a |
aCorrelation is significant at the 1% level.
Multicollinearity test coefficient table
| Collinearity statistics | ||
|---|---|---|
| Tolerance | VIF | |
| Total investment in fixed asset | 0.349 | 2.868 |
| Permanent population | 0.288 | 3.467 |
| Total water consumption | 0.360 | 2.776 |
| Collinearity statistics | ||
|---|---|---|
| Tolerance | VIF | |
| Total investment in fixed asset | 0.349 | 2.868 |
| Permanent population | 0.288 | 3.467 |
| Total water consumption | 0.360 | 2.776 |
It is clear from the computation results in Table 2 that there is a large gap between the 19 provinces and cities in China in terms of regional GDP and total investment in fixed assets, and the degree of data dispersion is relatively high. The level of economic development in each region will have an indirect impact on local water resources management, and the construction of water resources infrastructure also depends on the level of the local economy, thus affecting the efficiency of water resources utilization. Relatively speaking, there is a certain gap in the data in terms of permanent population and total water consumption, and the degree of dispersion is slightly lower.
According to the evaluation index screening principle, when measuring the efficiency with the DEA model, it is necessary to ensure the overall correlation between each evaluation index and avoid the problem of high collinearity among the evaluation indexes. From the Pearson correlation coefficient shown in Table 3, it can be seen that the three input indicators are positively correlated with the output indicators, and they are all significantly correlated at the 1% (two-tailed) level. This satisfies the requirement of inherent correlation between data when running the DEA model.
The test for multicollinearity mainly depends on the variance inflation factor (VIF). When the VIF ≥10, there is a high degree of collinearity among the variables, and the model construction is unreasonable. When the 0 < VIF < 10, there is no multicollinearity problem among the variables, and the model construction is reasonable. From the test coefficients shown in Table 4, it can be seen that the tolerances of the respective variables are greater than 0.1 and less than 1, and at the same time, the VIFs are far less than 10. Therefore, there is no multicollinear correlation between the variables, which meets the requirement that there is no high degree of collinearity among the data when running the DEA model. The efficiency evaluation index system constructed is reasonable.
RESULTS AND ANALYSIS
Calculation results of water resource utilization efficiency
Changes in water resource utilization efficiency before and after the operation of the water rights market under the BCC model in the first stage
This paper examines the water resources utilization efficiency of 19 Chinese provinces and cities using the input-oriented BCC model in Deap 2.1 efficiency calculation software (Wu et al. 2015). Table 6 shows the efficiency values and corresponding rankings of each province and city obtained in the first stage.
Water resources utilization efficiency before and after the operation of the WRs market under the BCC model in the first stage
| Province | Before the operation of the WRs market | After the operation of the WRs market | ||||
|---|---|---|---|---|---|---|
| y | PTE | SE | y | PTE | SE | |
| Tianjin | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Inner Mongoliaa | 0.685 | 0.700 | 0.980 | 0.619 | 0.650 | 0.949 |
| Heilongjiang | 0.430 | 0.467 | 0.921 | 0.366 | 0.482 | 0.760 |
| Shanghai | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Jiangsu | 0.757 | 1.000 | 0.757 | 0.829 | 1.000 | 0.829 |
| Anhui | 0.333 | 0.343 | 0.972 | 0.342 | 0.344 | 0.996 |
| Jiangxia | 0.335 | 0.370 | 0.906 | 0.348 | 0.372 | 0.934 |
| Henana | 0.558 | 0.798 | 0.699 | 0.479 | 0.791 | 0.606 |
| Hubeia | 0.450 | 0.472 | 0.954 | 0.476 | 0.506 | 0.941 |
| Guangdonga | 0.808 | 1.000 | 0.808 | 0.666 | 1.000 | 0.666 |
| Guangxi | 0.365 | 0.411 | 0.888 | 0.324 | 0.352 | 0.921 |
| Hainan | 0.406 | 0.945 | 0.431 | 0.385 | 0.965 | 0.399 |
| Chongqing | 0.538 | 0.598 | 0.899 | 0.546 | 0.583 | 0.936 |
| Sichuan | 0.484 | 0.551 | 0.879 | 0.412 | 0.509 | 0.810 |
| Guizhou | 0.392 | 0.543 | 0.722 | 0.357 | 0.467 | 0.765 |
| Yunnan | 0.383 | 0.456 | 0.840 | 0.326 | 0.388 | 0.839 |
| Gansua | 0.318 | 0.511 | 0.619 | 0.265 | 0.510 | 0.537 |
| Qinghai | 0.381 | 1.000 | 0.381 | 0.375 | 1.000 | 0.375 |
| Ningxiaa | 0.399 | 0.948 | 0.421 | 0.407 | 0.957 | 0.426 |
| Province | Before the operation of the WRs market | After the operation of the WRs market | ||||
|---|---|---|---|---|---|---|
| y | PTE | SE | y | PTE | SE | |
| Tianjin | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Inner Mongoliaa | 0.685 | 0.700 | 0.980 | 0.619 | 0.650 | 0.949 |
| Heilongjiang | 0.430 | 0.467 | 0.921 | 0.366 | 0.482 | 0.760 |
| Shanghai | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Jiangsu | 0.757 | 1.000 | 0.757 | 0.829 | 1.000 | 0.829 |
| Anhui | 0.333 | 0.343 | 0.972 | 0.342 | 0.344 | 0.996 |
| Jiangxia | 0.335 | 0.370 | 0.906 | 0.348 | 0.372 | 0.934 |
| Henana | 0.558 | 0.798 | 0.699 | 0.479 | 0.791 | 0.606 |
| Hubeia | 0.450 | 0.472 | 0.954 | 0.476 | 0.506 | 0.941 |
| Guangdonga | 0.808 | 1.000 | 0.808 | 0.666 | 1.000 | 0.666 |
| Guangxi | 0.365 | 0.411 | 0.888 | 0.324 | 0.352 | 0.921 |
| Hainan | 0.406 | 0.945 | 0.431 | 0.385 | 0.965 | 0.399 |
| Chongqing | 0.538 | 0.598 | 0.899 | 0.546 | 0.583 | 0.936 |
| Sichuan | 0.484 | 0.551 | 0.879 | 0.412 | 0.509 | 0.810 |
| Guizhou | 0.392 | 0.543 | 0.722 | 0.357 | 0.467 | 0.765 |
| Yunnan | 0.383 | 0.456 | 0.840 | 0.326 | 0.388 | 0.839 |
| Gansua | 0.318 | 0.511 | 0.619 | 0.265 | 0.510 | 0.537 |
| Qinghai | 0.381 | 1.000 | 0.381 | 0.375 | 1.000 | 0.375 |
| Ningxiaa | 0.399 | 0.948 | 0.421 | 0.407 | 0.957 | 0.426 |
Notes: All data are rounded to three decimal places. The efficiency value of the WRs market before operation is the average value for the period 2010–2013, and the efficiency value after operation is the average value for the period 2014–2019.
aPilot provinces/regions.
Table 6 shows the efficiency measuring findings for 19 provinces and cities from 2010 to 2019. However, since the calculation results of the first stage did not consider the interference of environmental variables, only a simple preliminary analysis is made here, and the following conclusions are obtained.
Before and after the operation of the WRs trading market, the degree of change in water resources utilization efficiency in the seven national WRs pilot areas are different. Among them, Jiangxi, Hubei, and Ningxia provinces have improved efficiency, while water resources utilization efficiency has declined in other pilot provinces. In the pilot areas, about 42.86% of the provinces have improved in efficiency. The water resources utilization efficiency in the non-pilot areas showed a downward trend as a whole, among which only Jiangsu, Anhui, and Chongqing saw a slight increase in water resources utilization efficiency. In pilot locations, the pace of improvement in water resources utilization efficiency is substantially faster than in non-pilot areas. It is preliminarily speculated that the operation of the WRs trading market has a positive impact on efficiency.
Among the pilot provinces, the decreased efficiency of the provinces may be due to objective reasons such as inadequate policy implementation, resulting in insignificant effects of WRs trading policies. However, why did the water resources utilization efficiency of some non-pilot provinces increase after the operation of the WRs trading market? The reason is that Jiangsu, Anhui, and Chongqing, where the efficiency has improved, have relatively high levels of economic development. Jiangsu and Anhui are significant cities in the lower Yangtze River Economic Belt, and Chongqing, as one of the four centrally administered municipalities, has a self-evident level of economic development. It can be found that the level of economic development will affect the regional water resources utilization efficiency. Therefore, the interference of environmental variables will be eliminated in the second stage to measure and calculate a more objective efficiency value.
The second-stage SFA-like regression analysis
The relaxation results of the three input variables selected in this paper (fixed asset investment relaxation, resident population relaxation, and total water consumption relaxation) are taken as the dependent variables of the equation. The selected three environmental variables (industrial structure, water resources endowment, and economic development level) are taken as independent variables of the equation. Here, the regression method of estimating 24 separate SFA equations is adopted, so that environmental variables will have different effects on different relaxations, which maintains the flexibility of the model and makes the efficiency measurement more effective. The regression model was established using Frontier software (Wang et al. 2019b), and the SFA regression results were obtained. Table 7 only shows the results for 2016.
SFA regression results in 2016 in the second stage
| Dependent variable/independent variable | Fixed asset investment relaxation | Resident population relaxation | Total water consumption relaxation |
|---|---|---|---|
| Constant term | 0.188 × 105a (0.156) | 0.538 × 104b (0.300) | 0.522 × 102b (0.998) |
| Industrial structure | −0.315 × 103b (0.193) | −0.361 × 102c (0.557) | 0.342 × 10a (0.113) |
| Water resources endowment | −0.704 (0.524) | −0.199 (0.125) | 0.150 × 10−1c (0.151) |
| Economic development level | −0.203 (0.276) | −0.511 × 10−1c (0.103) | −0.850 × 10−3a (0.453) |
| σ2 | 0.571 × 108c (0.100) | 0.154 × 107 (0.100) | 0.137 × 105b (1.000) |
| γ | 0.675c (0.232) | 0.364c (0.361) | 1.000a (0.212 × 10−5) |
| value of log | −0.191 × 103 | −0.162 × 103 | −0.104 × 103 |
| value of LR | 0.257 | 0.638 × 10 | 0.849 × 10 |
| Dependent variable/independent variable | Fixed asset investment relaxation | Resident population relaxation | Total water consumption relaxation |
|---|---|---|---|
| Constant term | 0.188 × 105a (0.156) | 0.538 × 104b (0.300) | 0.522 × 102b (0.998) |
| Industrial structure | −0.315 × 103b (0.193) | −0.361 × 102c (0.557) | 0.342 × 10a (0.113) |
| Water resources endowment | −0.704 (0.524) | −0.199 (0.125) | 0.150 × 10−1c (0.151) |
| Economic development level | −0.203 (0.276) | −0.511 × 10−1c (0.103) | −0.850 × 10−3a (0.453) |
| σ2 | 0.571 × 108c (0.100) | 0.154 × 107 (0.100) | 0.137 × 105b (1.000) |
| γ | 0.675c (0.232) | 0.364c (0.361) | 1.000a (0.212 × 10−5) |
| value of log | −0.191 × 103 | −0.162 × 103 | −0.104 × 103 |
| value of LR | 0.257 | 0.638 × 10 | 0.849 × 10 |
Note: The brackets in the table are the standard deviations of the corresponding coefficients.
aSignificance levels of 1%.
bSignificance levels of 5%.
cSignificance levels of 10%.
Since this study relates to water resources utilization efficiency, only the specific relationship between the three environmental variables and the relaxation of total water use is analyzed in. According to the SFA regression results for 2016 shown in Table 6, the following can be seen:
(1) At the 1% level, the industrial structure and the relaxation of total water utilization are considerably positively connected, indicating that a higher proportion of the primary industry is not conducive to improving water resources utilization efficiency. Industrial structure is closely related to water resources utilization efficiency, but there is a gap in water consumption among different industries. Therefore, the objectives are to optimize the industrial structure of the region, establish and improve farmland water conservancy infrastructure projects, promote scientific water-saving technologies, explore the confirmation of water use rights for agricultural water users, and explore the reform of property rights for small farmland water conservancy facilities, to work to increase China's agricultural water consumption efficiency.
(2) The water resources endowment and the relaxation of total water consumption are significantly positively correlated at the level of 10%, indicating that in the case of abundant water resources, attention should be paid to improving local awareness of water conservation. The endowment of water resources here is measured by the amount of water resources per capita. In general, economically developed areas have obvious population agglomeration and low endowment of water resources, but their water resources can be fully utilized, which is also in line with realistic logic. In areas with high water resources endowment, water users have insufficient awareness of water conservation, coupled with a lack of institutional constraints and technical support, resulting in low water resources utilization efficiency. Therefore, the water administrative department needs to accelerate the commercialization of water resources, steadily promote the WRs market, and use the market price mechanism to cultivate and improve the water-saving awareness of enterprises, residents, and other water users, so that water resources can be more fully utilized.
(3) At the 1% level, the level of economic development and the relaxation of total water use are considerably inversely associated, indicating that the more developed a region is, the more capital it has to utilize water resources reasonably and fully. On the one hand, the stable financial revenue of the regional government provides a financial guarantee for the construction of water resources utilization infrastructure and water pollution prevention and control facilities, thus providing technical support for the improvement of water resources utilization efficiency. On the other hand, the higher the degree of industrial agglomeration, the more developed the region, the higher the water resources utilization efficiency in the production process, and the scale effect of water resources utilization is obvious.
According to the above analysis, environmental variables will indeed interfere with efficiency calculations. Therefore, it is considered that the calculation of water resources utilization efficiency will have deviations due to the existence of environmental variables, and the efficiency calculation is carried out step by step under the condition of eliminating the interference of environmental variables.
Changes in water resource utilization efficiency before and after the operation of the water rights market under the BCC model in the third stage
In this part, the SFA regression results of the second stage operation are brought into Equation (3) to obtain the input value only related to management inefficiency, and then measure the water resources utilization efficiency again.
According to the calculation results in Table 8, the water resources utilization efficiency in the pilot areas increased before and after the operation of the WRs trading market. Among these areas, the water resources utilization efficiency of Henan, Gansu, and Ningxia decreased slightly. In the pilot areas, about 57.14% of the provinces improved in efficiency. The overall water resources utilization efficiency in non-pilot regions showed a downward trend, among which only Anhui, Guizhou, and Qinghai saw a slight increase in water resources utilization efficiency. The rate of improvement in water resources utilization efficiency in pilot areas was much faster than that in non-pilot areas. Therefore, the analysis of the reasons for the changes in water resources utilization efficiency in non-pilot provinces is limited in this paper, and the reasons for the decline in water resources utilization efficiency in Henan, Gansu, and Ningxia after the operation of the WRs trading market are also analyzed.
Henan is mainly engaged in WRs transactions for cross-basin water diversion projects. The completed water volume transactions between Pingdingshan and Xinmi, Nanyang and Dengfeng, and Nanyang and Xinzheng are of demonstration significance. However, the benefits of water trading in inter-basin water transfer projects are still limited and constrained by some factors:
(1) Although Henan has issued a series of policy documents to ensure the smooth implementation of water volume transactions for inter-basin water transfer projects, they do not have the same mandatory binding force as national policies.
(2) There is a problem of inconsistency between the construction of supporting projects and the WRs transaction, and the agreement reached will be affected by the construction period.
(3) Since water resources are the basic resources to guarantee people's life and economic development, in the initial stage of the WRs trading market, to ensure the stability of the market, the price of WRs trading would be dominated by policies and could not fully play the role of the market.
(4) Although the state proposes ‘three red lines’ for water resources, some entities still want to obtain project water quotas at no cost or at a low price, which is not conducive to the practice and promotion of water volume trading in water transfer projects.
(5) Engineering technical conditions restrict the trade of surplus water use indicators between regions.
The existence of this series of problems led to a slight decline in the efficiency of water resources utilization in Henan after the operation of the WRs trading market.
There are two reasons for the decline in water resources utilization efficiency in Gansu and Ningxia after the WRs trading market was launched:
(1) Gansu and Ningxia are mainly engaged in WRs transactions between water users, because the pace of industrial and service development in these two regions is relatively slow compared to other provinces. Therefore, the transaction volume of WRs transferred from agriculture to industry and service industry is not large. However, it is a ‘gentleman's agreement’ between farmers to borrow WRs as a daily use item, which has little impact on water resources utilization efficiency.
(2) Agricultural water utilization in Gansu and Ningxia accounts for more than 72% of the overall water consumption, and industrial water utilization accounts for less than 8% of the total water consumption. The development of industry and agriculture is not coordinated, and the agricultural water resources utilization efficiency is generally low, which is an important reason for the decline in its water resources utilization efficiency.
Policy effect test based on PSM-DID method
Analysis of DID results
Using the PSM approach to match the treatment and control groups, the suitable treatment group and control group were selected and the DID method was employed to obtain the impact of WRs trading policies on water resources utilization efficiency. Table 9 displays the regression results. When the control variable is not added, the double-difference regression coefficient is 0.041, and it is positively significant at the 10% significance level. After the control variables are introduced, the fitting coefficient R2 of the model increases from 0.291 to 0.487, and the double-difference regression coefficient is 0.027, which is positively significant at the 5% significance level. That is, regardless of whether control variables are considered, the DID regression coefficient is significantly positive, which shows that, compared to non-pilot locations, the functioning of the WRs trading market in pilot areas can improve water resources utilization efficiency.
Water resources utilization efficiency before and after the operation of the WRs market under the BCC model in the third stage
| Province | Before the operation of the WRs market | After the operation of the WRs market | ||||
|---|---|---|---|---|---|---|
| y | PTE | SE | y | PTE | SE | |
| Tianjin | 1.000 | 1.000 | 1.000 | 0.824 | 1.000 | 0.824 |
| Inner Mongolia* | 0.607 | 1.000 | 0.607 | 0.630 | 1.000 | 0.630 |
| Heilongjiang | 0.511 | 0.999 | 0.512 | 0.448 | 0.997 | 0.450 |
| Shanghai | 0.945 | 1.000 | 0.945 | 0.878 | 1.000 | 0.878 |
| Jiangsu | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Anhui | 0.608 | 0.997 | 0.609 | 0.612 | 0.998 | 0.613 |
| Jiangxi* | 0.488 | 0.997 | 0.490 | 0.497 | 0.994 | 0.500 |
| Henan* | 0.959 | 0.978 | 0.980 | 0.917 | 0.979 | 0.938 |
| Hubei* | 0.721 | 0.993 | 0.726 | 0.731 | 0.995 | 0.734 |
| Guangdong* | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Guangxi | 0.491 | 0.997 | 0.493 | 0.488 | 0.993 | 0.492 |
| Hainan | 0.274 | 1.000 | 0.274 | 0.239 | 1.000 | 0.239 |
| Chongqing | 0.690 | 1.000 | 0.690 | 0.683 | 1.000 | 0.683 |
| Sichuan | 0.756 | 0.933 | 0.810 | 0.735 | 0.950 | 0.775 |
| Guizhou | 0.427 | 0.999 | 0.428 | 0.469 | 0.998 | 0.470 |
| Yunnan | 0.476 | 0.996 | 0.478 | 0.474 | 0.995 | 0.477 |
| Gansu* | 0.331 | 1.000 | 0.331 | 0.298 | 0.999 | 0.298 |
| Qinghai | 0.123 | 1.000 | 0.123 | 0.137 | 1.000 | 0.137 |
| Ningxia* | 0.187 | 1.000 | 0.187 | 0.160 | 1.000 | 0.160 |
| Province | Before the operation of the WRs market | After the operation of the WRs market | ||||
|---|---|---|---|---|---|---|
| y | PTE | SE | y | PTE | SE | |
| Tianjin | 1.000 | 1.000 | 1.000 | 0.824 | 1.000 | 0.824 |
| Inner Mongolia* | 0.607 | 1.000 | 0.607 | 0.630 | 1.000 | 0.630 |
| Heilongjiang | 0.511 | 0.999 | 0.512 | 0.448 | 0.997 | 0.450 |
| Shanghai | 0.945 | 1.000 | 0.945 | 0.878 | 1.000 | 0.878 |
| Jiangsu | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Anhui | 0.608 | 0.997 | 0.609 | 0.612 | 0.998 | 0.613 |
| Jiangxi* | 0.488 | 0.997 | 0.490 | 0.497 | 0.994 | 0.500 |
| Henan* | 0.959 | 0.978 | 0.980 | 0.917 | 0.979 | 0.938 |
| Hubei* | 0.721 | 0.993 | 0.726 | 0.731 | 0.995 | 0.734 |
| Guangdong* | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |
| Guangxi | 0.491 | 0.997 | 0.493 | 0.488 | 0.993 | 0.492 |
| Hainan | 0.274 | 1.000 | 0.274 | 0.239 | 1.000 | 0.239 |
| Chongqing | 0.690 | 1.000 | 0.690 | 0.683 | 1.000 | 0.683 |
| Sichuan | 0.756 | 0.933 | 0.810 | 0.735 | 0.950 | 0.775 |
| Guizhou | 0.427 | 0.999 | 0.428 | 0.469 | 0.998 | 0.470 |
| Yunnan | 0.476 | 0.996 | 0.478 | 0.474 | 0.995 | 0.477 |
| Gansu* | 0.331 | 1.000 | 0.331 | 0.298 | 0.999 | 0.298 |
| Qinghai | 0.123 | 1.000 | 0.123 | 0.137 | 1.000 | 0.137 |
| Ningxia* | 0.187 | 1.000 | 0.187 | 0.160 | 1.000 | 0.160 |
* Pilot provinces/regions
Regression results of the impact of WRs trading on water resources utilization efficiency
| (1) y | (2) y | |
|---|---|---|
 | 0.041a (1.772) | 0.027b (2.144) |
| ln(x1) | −0.172 (−0.744) | |
| ln(x2) | −0.250 (−1.128) | |
| ln(x3) | 0.684a (1.841) | |
| ln(x4) | 0.071 (1.265) | |
| ln(x5) | −0.313 (−0.526) | |
| ln(x6) | −0.021c (−4.365) | |
| ln(x7) | 0.032 (0.835) | |
| ln(x8) | 0.026 (0.926) | |
| _cons | 0.631c (59.146) | −0.315 (−0.214) |
| N | 122 | 122 |
| R2 | 0.291 | 0.487 |
| (1) y | (2) y | |
|---|---|---|
| | 0.041a (1.772) | 0.027b (2.144) |
| ln(x1) | −0.172 (−0.744) | |
| ln(x2) | −0.250 (−1.128) | |
| ln(x3) | 0.684a (1.841) | |
| ln(x4) | 0.071 (1.265) | |
| ln(x5) | −0.313 (−0.526) | |
| ln(x6) | −0.021c (−4.365) | |
| ln(x7) | 0.032 (0.835) | |
| ln(x8) | 0.026 (0.926) | |
| _cons | 0.631c (59.146) | −0.315 (−0.214) |
| N | 122 | 122 |
| R2 | 0.291 | 0.487 |
Note: The values in brackets are t values.
aCoefficient estimates are significant at the 10% level.
bCoefficient estimates are significant at the 5% level.
cCoefficient estimates are significant at the 1% level.
Moreover, among the eight control variables, the secondary industry's proportion and the tertiary industry's proportion are negatively correlated with water resources utilization efficiency. This may be because the benefits of cross-industry transfer of water resources were not obvious at the initial stage of industrial transformation, and with the continuous expansion of the production scale of the secondary industry and the tertiary industry, the industrial water consumption has increased significantly. The urbanization rate, population density, and water resources utilization efficiency are positively correlated. This demonstrates that the continuing expansion in urban population will boost the obvious factor agglomeration impact, which is beneficial to the improvement of water resources utilization efficiency. Since this study used the average years of education to measure the quality of the population, and the level of education cannot completely determine the water-saving awareness of water users, there is a negative correlation between the quality of the population and the efficiency of water resources utilization. The proportion of GDP accounted for by investment in fixed assets for sewage treatment, area of water-saving irrigation, and investment in environmental pollution control represents the intensity of investment in water-saving measures by local governments. Looking at the regression coefficient as a whole, it can be concluded that the local government's investment in water conservation and water control technology and water resources infrastructure construction can positively affect the efficiency of water resources utilization.
Robustness test
(1) Parallel trend test. Validating the results of the DID estimation requires ensuring that the treatment and control groups follow a parallel trend assumption (Bertrand & Mullainathan 2004). If there is a time trend difference between the treatment group and the control group before the policy implementation, it may be suggested that the difference between the two groups after the policy implementation was not caused by the policy itself, but was due to the different time trends before the policy implementation (Yu & Zhang 2017). The parallel trend test was performed on the water resources utilization efficiency of the treatment group and the control group. As shown in Figure 2, before the introduction of the WRs trading policy, the water resources utilization efficiency of both groups followed a similar trend. However, after the enactment of the WRs trading policy, there was a noticeable divergence in the water use efficiency between the treatment group and the control group. As a result, the DID model was employed to investigate the impact of WRs trade regulations on water resources utilization efficiency, satisfying the parallel trend assumption criterion.
(2) Placebo test (that is, counterfactual testing, inferring the impact of policies or events by making assumptions that are contrary to the facts). This article assumes that there is an economic trend that leads to a false regression, then changing the time when the national WRs pilot is approved, the relevant conclusions can still be significant, otherwise there will be no economic growth trend. To advance the approval time of the seven national WRs pilots by 1 year. If the DID regression coefficient with water resources utilization efficiency is not significant, this proves that the improvement of water resources utilization efficiency in the pilot area is indeed caused by the national-level WRs pilot policy. The regression results are shown in Table 10. When the policy implementation time is projected forward by 1 year, the double-difference regression coefficient obtained from the operation is −0.0244, which is not statistically significant. This demonstrates that the improved water resources utilization efficiency in the pilot areas is linked to the 2014 introduction of the WRs trading policy.
Parallel trend chart of water resources utilization efficiency before and after the implementation of the WRs trading policy.
Parallel trend chart of water resources utilization efficiency before and after the implementation of the WRs trading policy.
Placebo test
| y | |
|---|---|
| Z | 0.0004 (0.0272) |
 before | −0.0244 (−1.3031) |
| ln(x1) | −0.0763 (−0.3994) |
| ln(x2) | −0.2181 (−1.1274) |
| ln(x3) | 0.5617 (1.6977) |
| ln(x4) | 0.0748 (1.4852) |
| ln(x5) | −0.3289 (−0.6904) |
| ln(x6) | −0.0184*** (−3.3028) |
| ln(x7) | 0.0363 (0.9979) |
| ln(x8) | 0.0144 (0.9937) |
| _cons | −0.3654 (−0.3352) |
| N | 139 |
| R2 | 0.4461 |
| y | |
|---|---|
| Z | 0.0004 (0.0272) |
| before | −0.0244 (−1.3031) |
| ln(x1) | −0.0763 (−0.3994) |
| ln(x2) | −0.2181 (−1.1274) |
| ln(x3) | 0.5617 (1.6977) |
| ln(x4) | 0.0748 (1.4852) |
| ln(x5) | −0.3289 (−0.6904) |
| ln(x6) | −0.0184*** (−3.3028) |
| ln(x7) | 0.0363 (0.9979) |
| ln(x8) | 0.0144 (0.9937) |
| _cons | −0.3654 (−0.3352) |
| N | 139 |
| R2 | 0.4461 |
*** Coefficient estimates are significant at the 1% level, as in Table 9.
DISCUSSION
WRs trading is a market-oriented behavior. Through macro-control and market mechanisms, strengthening the awareness of water conservation and scientifically allocating water resources will help improve the efficiency of water resources utilization. Compared with the traditional regression analysis method, the PSM-DID method has advantages in solving endogeneity problems, flexibility, and sample selection bias. The PSM-DID method has been widely used in many fields, such as technological innovation (Zhang & Gan 2023), economic development (Guo et al. 2023), enterprise management (Wu et al. 2023), and environmental governance (Yang et al. 2022), but there are few studies on the relationship between WRs and water resources using this method.
Our research validates how WRs trading policies impact water resources utilization efficiency via the PSM-DID approach. It is discovered that after the introduction of the WRs trading policy, the water resources utilization efficiency in most experimental areas increased due to the more adaptable allocation of water resources and the improved responsiveness, and there was a general upward trend. Among them, the decline in water resources utilization efficiency in Henan, Gansu, and Ningxia provinces, through the analysis of the actual development of these three provinces, further verified the impact of industrial structure and policy promotion on water resources efficiency:
(1) In provinces dominated by the primary industry, WRs transactions usually occur between farmers. Rigid policies at the national level are not binding enough on end users, and it is necessary to further refine WRs trading policies and expand the scope of application of WRs trading policies.
(2) The parallel trend test revealed a significant difference between the pilot and non-pilot areas following the policy. The non-pilot areas show a downward trend, while the pilot areas show an upward trend. This also shows that the optimization of water resources allocation and the enhancement of supply chain response capabilities brought about by WRs policies can create a better development environment for the improvement of water resources utilization efficiency. It is also conducive to further optimizing the transfer mechanism of WRs between regions, reducing the cost of water for the secondary and tertiary industries, increasing economic development and industrial transformation, and realizing the mutual promotion of water resources utilization efficiency-industrial transformation-economic development.
However, some scholars hold the view that China's WRs trading cases are confined to a few pilot projects (Sun et al. 2016), and the conversion of WRs between agriculture and industry is not prevalent (Wang et al. 2020). Hence, they contend that China's WRs trading is not successful (Moore & Yu 2020). Our findings indicate a different conclusion. As demonstrated by Xu et al. (2023), the concept of the ‘two hands’ approach is indeed a new perspective in discussing the relationship between the state and the market (Xu et al. 2023). The WRs trading policy in China is precisely a product of the ‘two hands’ approach. Practice has demonstrated that the implementation of WRs trading policy can indeed enhance the water resources utilization efficiency in pilot areas. According to data released by the China Water Rights Exchange, WRs trading activities have been conducted in most regions. As of July 2024, a total of 11,861 WRs trading projects have been promoted nationwide, with a total volume of water traded exceeding 454 million m3 and a transaction value reaching up to 2.6 billion yuan, as shown in Table 11. This indicates the work on WRs trading tends to be successful.
Current situation of WRs trading in China
| Transaction type | Total singular | Total water volume (104 m3) | Total amount (104 yuan) |
|---|---|---|---|
| Regional WRs | 41 | 129,850.00 | 88,181.51 |
| Water-drawing rights | 1,249 | 309,103.12 | 171,060.90 |
| Irrigation WRs | 10,569 | 15,321.53 | 1,660.11 |
| WRs of pipe network users | 2 | 3.30 | 0.42 |
| Total | 11,861 | 454,277.95 | 260,902.94 |
| Transaction type | Total singular | Total water volume (104 m3) | Total amount (104 yuan) |
|---|---|---|---|
| Regional WRs | 41 | 129,850.00 | 88,181.51 |
| Water-drawing rights | 1,249 | 309,103.12 | 171,060.90 |
| Irrigation WRs | 10,569 | 15,321.53 | 1,660.11 |
| WRs of pipe network users | 2 | 3.30 | 0.42 |
| Total | 11,861 | 454,277.95 | 260,902.94 |
Note: Data from the official website of China Water Rights Exchange.
In reality of the water shortages in China, the vigorous development of the WRs trading market is an inevitable result (Wang & Zhao 2022). In the scenario of strictly implementing the ‘ceiling’ of the total amount of water resources use, water users must explore the possibility of WRs trading. However, in the process of societal development, the market-oriented management mechanism cannot completely rely on government regulation. Compared with using marketization to stimulate the water-saving consciousness of water-using subjects, the institutional pressure formed by government management has further dispelled its endogenous motivation for water-saving. In addition, it is necessary for the government to create a dynamic water-saving environment through moderate and humanized policy design, and provide institutional guarantees and behavioral supervision for WRs transactions.
CONCLUSIONS
In this study, a three-stage DEA method was initially employed to estimate the water consumption efficiency of 19 Chinese provinces and cities from 2010 to 2019, while accounting for the effects of removing external environmental factors and random noise. The primary findings are as follows, based on the use of PSM-DID to analyze the effect of WRs trading policy on water resources utilization efficiency:
(1) According to the calculation findings of the three-stage DEA model, the water resources utilization efficiency in the pilot regions improved compared to the non-pilot areas after the adoption of the WRs trading policy.
(2) The research results based on the PSM-DID method show that the WRs trading policy has had a positive effect on the improvement of water resources utilization efficiency.
(3) Furthermore, the results also indicate that the share of secondary industry, the share of tertiary industry, population quality, and water use efficiency have a negative correlation. It can also be observed that the urbanization rate, population density, fixed asset investment in sewage treatment, water-saving irrigation area, and the ratio of environmental pollution control investment to GDP have a positive association with water resources utilization efficiency.
Based on the empirical study presented above, it is suggested that China should continue to investigate and develop the WRs trading system. Since the operation of the WRs trading market, the efficiency of water resources utilization in each pilot area has basically improved, but needs to be improved further. WRs trading is a typical application of market-based means in water resources allocation. The market plays a decisive role in improving the efficiency of resource allocation, optimizing the structure of water use, and promoting the development and application of water-saving technologies. Therefore, we should learn from the practical experience of the pilot areas, continuously improve the WRs trading system, and promote the operation of the national WRs trading market according to local conditions, according to different economic development levels and water resources conditions in different regions.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.
%20cropped.png?versionId=7153)